Event detection using the user context in sensor based IoT

The clustering methodologies for deployed devices in the region of interest consider the similarity of sensed data or target maximization of network lifetime. Thus the clusters formed are either sensing clusters or communication clusters. In the case of motes capable of sensing multi-feature data, the similarity of sensed parameters is generally used to cluster the devices. Event detection methodologies in uni-modal sensor based IoT mainly focus on communication clusters to maximize network lifetime. Yet, the existing approaches for clustered data gathering hardly incorporate user context and preferences for sensing cluster formation and event detection. Additionally, many approaches do not incorporate both sensing and communication clusters in order to detect the event and balance it with energy efficiency. Research work that considers user context in the domain of the internet of things is also limited with a focus on network lifetime, event detection accuracy etc. The present work aims at resolving the existing constraints in 'user context' aware clustering methods and event detection. The user context parameters based on the domain knowledge of the user are used to divide the deployed region into sub-regions forming sensing clusters. The methodology allows the user to change the context and definition of events for each sub-region for accurate, context-aware, and discriminatory event detection. Additionally, the compressive gathering of detected events results in energy efficient data transmission. Simulations were performed for scalability and comparative analysis of the proposed scheme. The results show that the proposed scheme outperforms the existing schemes in terms of detection accuracy and network lifetime.